Suggested Methodology of Building BI Systems

Building and implementing BI systems require organisations to have some culture of working
with information and information technologies, which is related to:

thorough and ongoing research into organisations informational needs (present and future);

authentic co-operation of the users involved (i.e. decision makers and operational personnel) with organisations IT departments and knowledge management centres;

information sharing; and

abilities to interpret analyses and use such analyses in management properly.

Suggesting the methodology of building and implementing BI systems, the authors have benefited from sound business practices set by enterprises that succeed in building BI systems. Any
general model to be suggested may be a set of guidelines and some kind of a pattern for organisations that want to use any BI system. The model in question pays particular attention to the role of
end users in the whole life cycle of the BI systems including the phase of the BI usage in particular. Parameterisation of the BI system carried out by its user is required for its correct performance. Such parameterisation involves providing repositories with knowledge (employees, customers, suppliers or co-operators). Using BI systems will succeed in business only when their
users keep:

identifying and modelling knowledge;

monitoring and modifying data repositories;

creating their own analyses and reports;

learning how to interpret results and ask sophisticated questions; and

improving business and decision making on the ongoing basis.

All the above has to be manifested in the system performance. Taking into account significant
influence users have on the BI system performance allows for suggesting two major iterative
stages in the approach to building and creating the systems in question, i.e. (compare: Dresner,
Buytendijk, Linden, Friedman, Strange, Knox, & Camn, 2002):

139

Building and Implementing Business Intelligence Systems

creation of BI; and

use (consumption) of BI.

BI creation is the most time consuming and this stage requires the greatest part of financial and
manpower resources in the whole BI life cycle. BI creation consists of numerous stages including
in particular:

definition of the BI undertaking, i.e. determination of the BI system development strategies;

identification and preparation of source data;

selection of BI tools;

designing and implementing of BI; and

discovering and exploring new informational needs and other business applications and
practices.

The BI consumption stage is predominantly associated with end user application. The stage
shows its major role in popularising and promoting practices that are related to data analyses and
BI systems. This stage may be divided into several different steps to be taken at the discretion of
users and according to their needs or tasks to be faced. The steps mainly include the following:

logistic analyses that enable to identify partners of supply chain quickly;

access, monitoring and analyses of facts;

development of alternative decisions;

division and co-operation; and

change in the effect of company performance.

Strategy of the BI System Development

Undertaking realisation of BI systems in organisations, it is first necessary to determine a general
vision of such systems. The systems also have to be related to business objectives. This stage involves specifying informational needs of organisations, simultaneously paying attention to key IT
decision makers and specialists. Ranking of informational needs (carried out on the basis of their
importance) allows for highlighting e.g. indexes that are important while realising business
strategies (Chaudhary, 2004).
It is necessary to remember that applying of BI systems in organisations makes sense when companies involved are not interested in passive recording of different events only. Such companies
ought to focus their attention on interpretation of different events in different wider contexts, e.g.
when it is more important to understand customers expectations and preferences than to find out
the amount of income obtained. BI systems should closely correspond to business objectives of
enterprises. Therefore, the most important motives that support implementation of BI systems in
enterprises may include the following (Kalakota, & Robinson, 1999; Liautaud, & Hammond,
2002; Rasmussen, Goldy, & Solli, 2002):

140

transitioning from instinct and intuition decision making to objectivism that is based on
the analysis of facts, indexes, balanced score cards, managerial cockpits, etc.;

forecasting enterprise development along with customers and suppliers behaviour;

matching operational activities with realisation of strategic objectives (measuring development in the realisation of strategies, monitoring of business process effectiveness,
matching budgets and investments with corporate strategies);

implementing standards that are used as the basis for repetitive, regular and cyclical business processes within organisations;

unifying informational transfers in order to make them more transparent and unifying
roles of individuals who participate in decision making processes;

rapid detecting of information that deviates from commonly accepted standards and procedures and that suggests some possibilities that new threats will emerge (dishonest customers, inflated material or energy usage, etc.);

shortening time that is necessary to analyse information, and decreasing a number of participants who are involved in analysing and processing of information; and

automatic and rapid reporting and preparing of plans and forecasts.

At this stage, it is necessary to find areas and business processes that will first undergo different
analyses and explorations. Analysing this process in enterprises that have already implemented BI
systems, it is possible to note that it was natural for the enterprises in question to start with finance. Then, marketing, customer relations management and logistics followed. As far as sectors
are concerned, enterprises that pioneered BI implementation mainly include banks and insurance
institutions that were subsequently followed by telecommunications, power engineering, transportation, trade companies and production enterprises.
Observing general trends in companies that show BI initiatives, it is possible to assume that the
development of the majority of BI is characterised by a top-down approach, i.e. firstly, decision
making by the board and top management is supported and then lower levels of management are
involved. At this stage, it is necessary to decide whether a particular BI solution is planned to be
implemented in one department and if it is supposed to cover a selected scope of enterprise functioning or whether a potential solution is to be of complex nature, thus integrating different aspects of activities undertaken by enterprises. Translating this situation into the data warehouse
context may provide an answer to the question if a particular enterprise is going to create individual data marts (that are subsequently going to be integrated in a corporate data warehouse), or
whether the enterprise in question will immediately take up building an integrated data warehouse
that would cover the enterprises different departments. The data marts oriented approach is supported by a less demanding scope of a project and - as a result - by faster effects to be obtained.
However, it is possible to observe some future difficulties while integrating different individual
data marts in one corporate data warehouse.
Formulating general requirements for potential BI systems, it is also worth considering whether,
for instance, such systems should work online, whether they ought to be directed at symmetric
exploration of Internet websites or at advanced forecasting analyses, etc. At this stage, available
sources of information are not taken into consideration. The objective is rather to identify general
business needs, even if it turns out that supporting such needs might be difficult or hardly possible.

Identification and Preparation of Data for the BI System

At the stage of identifying and preparing data, it is necessary to specify sources of data that may
be used to support business needs. Such specification requires finding internal sources (intellectual resources, information technology resources, paper files, etc.) and external ones (concentrated on customers, suppliers, shareholders, etc) along with verifying reliability of the sources

141

Building and Implementing Business Intelligence Systems

and a form of transformations that the sources in question will have to undergo so that they could
be subject to further analyses. Realisation of this stage calls for remarkable help provided by decision makers, operational workers, IT departments, departments of knowledge management and
strategic customers.
At this stage, it is necessary to diagnose all information systems and databases used in a given
enterprise (simple applications that are based on dbf files, Access or sophisticated ERP systems).
While carrying out such a diagnosis, it is worth following several important instructions (Botnicki, & Wawrzynek, 2006):

find data that are to be found in the IS and that are not important from the perspective of
the analyses carried out;

find relations between data in different information systems;

describe the logical structure of data that are to be found in the system: much attention
should be paid not only to their structure in the base but also to the relation with business
processes described;

find places that allow for generating errors in data (a possibility

that data are inconsistent); and

find limits of IS applicability (which data cannot be reported out of IS, e.g. due to erroneous logic of source bases).

At this stage, it may turn out that a lot of data are hidden or just inaccessible, and that it is necessary to use numerous informal sources. A huge amount of valuable data that are necessary for
analytic reports may be hidden, e.g. in the Internet resources. While searching for such data, it is
suggested to take advantage of the following methods:

manual searching: documents are directly downloaded by a particular system user;

automatic searching and downloading of HTML documents by means of appropriate

software; and

queries: using browsers that search through indexes.

Contents mined this way may provide interesting information. Processes of mining depend on the
data obtained. When data get filtered and are made homogenous, they may be directly transferred
to databases. ETL techniques are responsible for this process to the large extent. In order to mine
information from semistructured and unstructured sources, techniques of the so-called wrappers
along with text processing methods may prove useful (Poul, Gautman, & Balint, 2003).
Apart from text identification, it is recommended to carry out reliability, up-to-datedness, precision and consistency assessments. It is also necessary to find out whether data of the BI system
should be provided systematically or with some minimal delay or periodically.

Selection of BI Tools
Selection of a BI tool may turn out to be a difficult task. At present companies offer a wide range
of products beginning from simple reporting technologies up to sophisticated BI platforms. While
choosing a BI tool, it is necessary like in the case of purchasing other IS to take the following
criteria into consideration: functionality, complexity of solutions, and compatibility. It is also
necessary to remember that organisations informational needs will evolve. Therefore, BI tools
should be up-to-date enough to meet enterprises expectations in a few years to come.

142

At this stage, good market knowledge of BI is required. Today BI products may be found in different segments of the IT market (Ilczew, 2006). Providers of MRP II and ERP systems more
and more frequently equip their products with BI modules (e.g. SAP, Oracle or Microsoft), thus
wishing to make their products more dynamic and analytical. OLAP techniques and data mining
have also been implemented in database systems (Oracle, Microsoft or IBM). Planning and budgeting belong to another segment of the IT market that uses BI techniques. Additionally, it has to
be mentioned that there is a group of providers that offer BI solutions in a highly specialised area
and usually on a very high level of customer need satisfaction. Such products often include best
practices for a particular sector along with some future solutions. One cannot forget about open
source solutions that are more and more frequently available on the market.
In the BI sector similarly as in case of other IT sectors it is possible to observe some processes
of consolidating providers - purchasing products or expanding products by means of functionalities that are offered by the best providers in a given category. Hence, it is necessary to consider
whether a given enterprise ought to purchase products and technologies from one provider or if
such an enterprise should follow a principle of selecting the best products in a given category
(e.g. the best tools for OLAP, ETL, etc.) sold by different providers. In the former case, enterprises are guaranteed integration of particular products and a similar interface. However, it has to
be taken into account that not all solutions are going to be of the highest possible quality. Package
purchase of products frequently involves discounts, which is quite important for enterprises. On
the other hand, purchasing products from several providers may lead to delegating responsibility
for particular module performance to other providers. It is also more difficult to obtain larger discounts while purchasing technologies that come from different providers. There is also some
other possibility purchase of a ready to use solution instead of a particular technology. In this
situation, it is necessary to learn more about capacities of a given application and then consider
whether such an application meets enterprises needs and whether there are some elements that
the application in question should be subsequently provided with. Providing an enterprise with BI
products of an open source type is another possibility. Examples of complimentary or open
source products may be provided by Sygate Analyst (a tool used for data visualisation), Agata
Reports (a reporting tool), Oracle Application Express (environment for building web applications), and cockpit for the management in open source ERP Compiere, Business Intelligence Reporting Tool for Eclipse or Mondrian OLAP Server. Some providers of BI products use free databases. For instance, Business Objects uses a complimentary database called MySQL (Poul,
Gautman, & Balint, 2003).

Designing and Implementing the BI System

A scale of effort undertaken at the stage of designing and implementing BI varies. Such effort
predominantly depends on the system complexity and the level of its popularity. However, in the
majority of cases, creation of a customised BI application requires a lot of time. The time required
is spent not only on designing individual interfaces but also on making sure that the whole BI application is logical and consistent.
Another important stage of designing BI involves building a data warehouse that is supposed to
perform two functions: of a repository for further analyses, and of a base for the BI system (Inmon, 1992). This process has to be carried out in compliance with the following rules (Botnicki,
& Wawrzynek 2006; Hackathorn, 1998):

setting a scope of data stored in the IS that are important from a perspective of a given
organisation;

defining interconnections between data that are to be found in different systems and that
are of the same importance. As a result of such activities, a set of data will be created and

143

Building and Implementing Business Intelligence Systems

the data in question will allow for designing a target database (a repository) where data
from source bases will be sent; and

creating a design of a data warehouse that serves as a basis for loading a BI system. Such
a design should be created in order to provide easy configuration of database related reporting and querying mechanisms. The design is suggested to aim at reaching a model of
a star or snowflake that simplifies further implementation of data warehouse mechanisms including OLAP or data mining.

In order to make sure that a data warehouse is systematically updated to include data that come
from e.g. transaction systems, it is necessary to create mechanisms of data import (Meyer, 2001).
Such mechanisms ought to allow for importing all data and should enable users to perform incremental import that only requires processing of data that have reached source systems since the
last import. Incremental import does not burden source systems and mechanisms of data processing. Mechanisms of data transfer simultaneously perform a controlling function that is responsible for data consistency. In many cases the mechanisms in question allow for finding inconsistencies and errors at the stage of implementation. Import procedures are created so that they could
record erroneous data to be then transferred to a repository base, which enables to correct the data
in the place of their origination, and subsequently - to transfer such data to the base.
A created data warehouse ought to be the basis for building modules within reports and reporting.
It is necessary to provide for a few or at least two groups of reports. The former should cover predefined reports that are systematically updated by potential users. Such a group should also provide more advanced users with some possibilities to create their own reports according to their
individual needs. The latter type of reports and reporting involves reports that result from unexpected individual needs that concern details to be found in data.
Experience shows that - sooner or later - enterprises will need to use multidimensional analyses
(Rasmussen et al., 2002), which means that a BI solution ought to aim at OLAP modules (that let
users mine and view data in different perspectives) and data mining (that contributes to better understanding of customers preferences, nature of such preferences, supply chain, geographical
impacts, etc.) (Grossman, 1998; Kantardzic, 2002; Perkowitz, & Etzioni, 1999). Depending on
the specifics and needs of organisations, it is possible to consider carrying out analyses that are
responsible for, inter alia (Olszak, & Ziemba, 2006):

examining customers profitability and their lifetime value;

verifying importance of parameters;

segmenting and profiling of customers;

monitoring customers loyalty and withdrawals;

checking customers maintenance;

finding similarities;

studying frauds;

studying marketing campaigns;

and carrying out cross sales.

A layer of data presentation is an important element of BI systems. With reference to the above, it
is necessary to pay much attention to designing an interface. Development of interfaces is clearly
influenced by two tendencies. The former involves application of a spreadsheet as an interface.
For instance, Oracle provides business users of its Oracle 10g database with a possibility to use

144

the Microsoft Excel spreadsheet. The former tendency involves using internet browsers. In such
a situation, a BI application is not confined to users in a given enterprise only it may be
taken advantage of by external partners as well (Ilczew, 2006).

Exploring and Discovering New Informational Needs

The stage of exploration and discovery of new informational needs is of critical importance for
the whole cycle of building any BI. Implemented BI environment casts new light on the role of
information and competencies in an enterprise and on business relations and interdependencies.
At this stage, new informational needs emerge and new methods of information management are
created. Using notions of iterative designing and rapid application development (RAD), it is possible to observe that the BI environment obtained is a prototype that should serve as the basis for
evaluation and commencement of a new cycle of building a BI application (Dresner et al., 2002).
The exploration and discovery stage requires remarkable co-operation of representatives of IT
departments with a centre for knowledge management along with end users whose all innovativeness and willingness to experiment with data are of much importance.

BI consumption stage
Depending on tasks to be undertaken by users, a process of using different data repositories along
with ready to use analyses starts. Data are compared and interpreted. This stage requires users to
show much initiative. Depending on emerging needs, users should create author analyses and reports and be able to ask relevant questions and interpret results to be obtained.
Analyses of different facts may reveal alternative methods of solving a particular task and some
other possibilities of optimising numerous activities. A selection of a final decision to be made
frequently requires consultations with other employees and decision makers.
Analyses of the already mentioned questions and answers should cover co-operation and communication of all individuals who participate in decision making of an enterprise. Such behaviour
involves some necessity to eliminate problems related to, inter alia, privacy and data appropriation in an enterprise.
Work with information and tools used to analyse and explore data allows for carrying out fundamental changes in a decision making process. Changes in enterprise performance involves, inter
alia, looking for new forms of co-operation or outsourcing, new markets and business partners,
etc. (Linoff, & Berty, 2002; Srivastava, 2003).
Accomplishment of the last stage of the cycle of building and using BI systems does not mean
that all BI related problems are dealt with in a given enterprise. As it has already been mentioned,
the cycle in question is of iterative nature, thus being some kind of a loop that continuously requires carrying out more and more analyses of informational needs, re-evaluation of already existing solutions and their modifications, optimisations and adjustment.